Close this search box.

GUEST COMMENT Data-driven marketing: all in good time

This is an archived article - we have removed images and other assets but have left the text unchanged for your reference

Online retailers must struggle with a great irony – marketing technologies that identify users with buying intent and lead them to an eventual purchase can act as agents of delay, driving users away from a purchase.

In a study undertaken by Walmart that compared page load times to the likelihood of purchase, they found that a one second delay caused a 7% drop in conversions. This almost certainly mirrors conclusions from any online retail organisation – the longer it takes to make purchases, the fewer purchases a user is likely to make. Third-party site analytics tools, advertising pixels, and cookie syncing scripts can be a liability, particularly during times of increased traffic. Optimising page architecture can increase conversions and avoid catastrophe, but third-party servers are also under heavy loads. If these tools are required for page functionality, slow loads can be especially detrimental to the user experience. Even if latent tools don’t have an overt effect on a customer’s engagement, a failure to load in time can mean a failure to use data to target a user effectively in the future.

Site operators must attempt to determine to what degree these technologies contribute to the latency of their pages, but the relationship between data marketing technologies and page load speed isn’t always as intuitive as it seems. In fact, many sites manage their tags in a way that allows them to partner with a large number of marketing and analytics companies without suffering longer page load times.

We conducted a survey of the marketing activities and load times of 22 top UK retail sites, and compared the heaviest trafficked days of the Internet year (11/28-12/02) to the same five-day period one month before (10/31-11/04). The data shows wide and varying strategies for data collection and site optimisation, and it suggests equally varying success. (Note: the averages discussed in the following text refer to the entire 22 site study, while the graphs highlight the ten most relevant sites in that category.)

Unfortunately for some retailers, the removal of third-party code was not enough to overcome performance obstacles. discarded 82 distinct partners during the busy weekend, but that did not translate into quicker overall page times for the whole site (sites were fractionally slower on average by 166ms).

Other sites enjoyed success that, on the surface, may seem surprising. Consider the web properties for department store John Lewis. added code from 48 distinct partners, but improved its page load time by 260ms. Similarly, improved site performance by an average of ½ second despite adding 19 new partners; and added three new partners, but improved load times by nearly a full second.

Direct correlations do appear on certain sites. Travelocity’s improved by three-quarters of a second in page latency, perhaps due to the removal of 13 marketing partners. Tesco launched code from only four new companies and saw a similarly modest increase in page load times (190ms). But these types of connections are far from the norm. The top 20 sites were actually faster over the heavy shopping weekend, clocking an average overall improvement of 198 milliseconds (about fifth of a second faster), but that improvement didn’t come by eschewing third-party marketing tools. On average, top retail sites added six data collection technologies each. So better performance is not a simple matter of direct marketing austerity.

Through the use of marketing technologies, sites can gain a very clear understanding of their audience – but they can also complicate and confuse the operation of their own site. Retailers, like all website operators, must also conduct internal analysis so they can properly identify their most sound practices.

Online retail giant Amazon is notably successful at this is. Their site added 57 distinct partners when compared to the month before and those technologies were actually slightly slower to load on average. Amazon’s pages were slightly speedier overall, however, by a margin of 133 milliseconds, which was accomplished through particularly effective tag management. While scripts from 266 distinct partners were deployed across the domain throughout the weekend, relatively few are deployed on any given page view, and those are chosen via a sophisticated set of matching practices.

A retailer can judiciously choose which online marketing tools to deploy without automatically choosing to deploy fewer tools on the whole. Effective practices for measuring and monitoring the success of third-parties coupled with effective tag management can allow a site to reach a seemingly unattainable goal – aggressive adoption of partners to identify the right audience and lead them toward a purchase – but also give them a quick and seamless experience when they arrive.

Andy Kahl is director, data analysis, at Evidon.

Read More

Register for Newsletter

Group 4 Copy 3Created with Sketch.

Receive 3 newsletters per week

Group 3Created with Sketch.

Gain access to all Top500 research

Group 4Created with Sketch.

Personalise your experience on